On stop conditions about methods to obtain polygonal approximations relied on break point suppression

摘要

In this work, we propose a new unsupervised stop condition for heuristic methods that employ break point suppression to obtain polygonal approximations. Using a stop condition, these heuristic methods delete redundant break points until a required level of approximation is satisfied. The new stop condition is based on the optimisation of the F2 measure. Comparisons with original stop conditions of several methods show that the new stop condition yields a measurable improvement in the approximation. The polygonal approximations obtained with the new stop condition are more efficient and better adjusted to the original contours. The new stop condition can be combined with supervised optimal methods to obtain a reference approximation of the original contour with a minimum number of points. In this case, the resulting methods are unsupervised methods.